Primary pulmonary alveolar proteinosis (PAP) is a rare syndrome characterized by accumulation of surfactant in the lungs that is presumed to be mediated by disruption of granulocyte/macrophage colony-stimulating factor (GM-CSF) signaling based on studies in genetically modified mice. The effects of GM-CSF are mediated by heterologous receptors composed of GM-CSF binding (GM-CSF-Rα) and nonbinding affinity-enhancing (GM-CSF-Rβ) subunits. We describe PAP, failure to thrive, and increased GM-CSF levels in two sisters aged 6 and 8 yr with abnormalities of both GM-CSF-Rα–encoding alleles (CSF2RA). One was a 1.6-Mb deletion in the pseudoautosomal region of one maternal X chromosome encompassing CSF2RA. The other, a point mutation in the paternal X chromosome allele encoding a G174R substitution, altered an N-linked glycosylation site within the cytokine binding domain and glycosylation of GM-CSF-Rα, severely reducing GM-CSF binding, receptor signaling, and GM-CSF–dependent functions in primary myeloid cells. Transfection of cloned cDNAs faithfully reproduced the signaling defect at physiological GM-CSF concentrations. Interestingly, at high GM-CSF concentrations similar to those observed in the index patient, signaling was partially rescued, thereby providing a molecular explanation for the slow progression of disease in these children. These results establish that GM-CSF signaling is critical for surfactant homeostasis in humans and demonstrate that mutations in CSF2RA cause familial PAP.
Pulmonary surfactant homeostasis is maintained by the balanced production of surfactant by alveolar epithelium and its clearance by pulmonary alveolar macrophages (1). GM-CSF is a critical regulator of surfactant homeostasis in mice (1). GM-CSF, via heterologous cell surface receptors comprised of α and β glycoprotein subunits (GM-CSF-Rα and GM-CSF-Rβ, respectively), activates STAT5 and other signaling pathways (2). N-linked glycosylation of the α subunit is required for GM-CSF binding and receptor signaling (3). In mice, deficiency of GM-CSF (4, 5) or its receptor (6) impairs multiple functions of alveolar macrophages (7), including surfactant catabolism (8), which results in progressive surfactant accumulation.
Pulmonary alveolar proteinosis (PAP) is a syndrome characterized by respiratory failure caused by pulmonary surfactant accumulation (9, 10), which can be grouped into distinct categories based on clinical, histopathologic, biochemical, and genetic data (1, 9–12). Primary PAP occurs when GM-CSF signaling is disrupted, for example in individuals with high levels of GM-CSF autoantibodies, which are presumed to mediate pathogenesis by neutralizing GM-CSF and reducing surfactant catabolism by alveolar macrophages (10, 13). This form, referred to as autoimmune PAP, comprises 90% of cases (13). Primary PAP has also been associated with reduced detection of GM-CSF-Rβ on myeloid cells (14, 15), but definitive studies demonstrating heritable deficiency of either GM-CSF or its receptors as the cause of PAP in humans are lacking (16). Secondary PAP occurs as a consequence of an underlying disease presumed to impair surfactant clearance by reducing either the numbers or functions of alveolar macrophages (9). Hereditary disorders of surfactant production, for example, because of mutations in the genes encoding surfactant protein (SP)–B (17), SP-C (18), or ABCA3 (19), exhibit disordered surfactant homeostasis to varying degrees but are distinguished from PAP by their surfactant dysfunction, disruption of alveolar wall architecture, and clinical course (20). In this paper, we describe a family in which two children developed primary PAP in association with loss of GM-CSF responsiveness caused by mutations in the gene encoding GM-CSF-Rα.
Results And Discussion
Clinical presentation and phenotype
The index patient presented at age six with a 2-yr history of progressive tachypnea and failure to thrive (height and weight were third percentile for age [Fig. S1 A]). Gestation, delivery, birth weight (3.66 kg, 50th percentile), and development were all normal, but weight gain slowed by 6–9 mo as did height by 2–3 yr. There was no history of cough, fever, chest pain, pneumonia or other pulmonary disease, environmental exposure, or drug use. Both parents were well-developed and healthy with no history of lung disease. Examination revealed moderate tachypnea, mild tachycardia, and inspiratory crackles but was otherwise unremarkable. Pulmonary function testing revealed severe restrictive impairment. Oxygen saturation was 88% while breathing room air and decreased while talking or walking a short distance. A diagnosis of PAP was suspected based on chest radiography (Fig. 1 A) and established by histopathologic examination of lung tissue (Fig. 1 B). A serum GM-CSF autoantibody test (21) was negative on two occasions. The patient was transferred to Cincinnati Children's Hospital and underwent whole lung lavage therapy with marked symptomatic and radiographical improvement (Fig. 1 C). MCP-1 and M-CSF levels in the lavage were increased compared with healthy controls (supplemental text), which is similar to findings in autoimmune PAP patients and GM-CSF KO mice (22, 23).
An inherited defect in GM-CSF receptor function was suspected based on the constellation of histopathology showing well-preserved alveolar wall architecture (Fig. 1 B), the absence of GM-CSF autoantibodies (21), and an abnormal CD11b stimulation index test (24). Therefore, all immediate family members were evaluated. Because SP is in serum in PAP and levels reflect lung disease severity (25), serum SP-D was measured and found to be increased in the patient compared with her parents and controls (Fig. 1 D). Unexpectedly, the patient's 8-yr-old sister, who was previously thought to be healthy, had increased serum SP-D. Subsequent clinical evaluation revealed that the sister had poor growth (height and weight, 10th percentile for age; Fig. S1B), a diffusion capacity for carbon dioxide of 57% that predicted, and mild patchy ground glass opacities throughout both lungs, which are consistent with a diagnosis of PAP (Fig. 1 E).
Structure and function of the GM-CSF receptor
Flow cytometry indicated that GM-CSF-Rα and GM-CSF-Rβ were present on blood leukocytes in all family members (Fig. 2 A). Western blotting revealed an abnormal electrophoretic pattern for GM-CSF-Rα in the patient and her sister with predominantly lower molecular mass forms compared with the control (Fig. 2 B). The father was heterozygous for normal and lower molecular mass forms. Unexpectedly, the mother exhibited only a normal pattern. To evaluate cell-mediated clearance of GM-CSF, blood mononuclear cells from the patient (Fig. 2 C) and her sister (Fig. S2) were incubated with GM-CSF, and both failed to clear GM-CSF in contrast to control and parental leukocytes. The concentration of GM-CSF in lung lavage was markedly increased in the patient compared with healthy controls, in whom it was undetectable (Fig. 2 D). The in vivo concentration of pulmonary GM-CSF in the epithelial lining fluid of the lungs was estimated to be 6.9 and 27 ng/ml (right and left lungs, respectively) using the urea dilution method (26). Pulmonary GM-CSF was also increased in CSF2RB-deficient mice (Fig. S3), confirming the relationship between GM-CSF receptor dysfunction and increased pulmonary GM-CSF levels. Serum GM-CSF was increased in both affected siblings and undetectable in both parents and control, indicating that the abnormality was systemic (Fig. 2 E). GM-CSF receptor function was impaired in both affected siblings, as shown by reduced GM-CSF–stimulated phosphorylation of STAT5 (Fig. 2 F) and failure of GM-CSF to increase CD11b levels on leukocytes from the patient compared with controls (Fig. 2 G).
The slower-than-expected rate of progression of PAP in both affected siblings led us to hypothesize that the GM-CSF-Rα abnormality may not completely abolish GM-CSF signaling. This was confirmed by incubation of primary blood mononuclear cells with a higher concentration of GM-CSF (1,000 ng/ml), which resulted in partial receptor signaling as shown by an increase in phosphorylated STAT5 (Fig. 2 H). Thus, although both GM-CSF-Rα and β chains were present on leukocytes in all family members, in the two affected siblings, the GM-CSF-Rα chain was structurally abnormal, which severely reduced but did not completely abolish GM-CSF binding and GM-CSF–dependent signaling and cellular functions.
Genetic analysis of the GM-CSF receptor
The nucleotide sequence of the CSF2RA gene in genomic DNA from both siblings revealed a single G→A point mutation in exon 7 that encoded a nonconservative amino acid change (glycine to arginine) at position 174, G174R (Fig. 3 A). This change alters one of the 11 potential sites of N-glycosylation, which is required for GM-CSF receptor function (3, 27). The father was heterozygous for mutant CSF2RAG174R and WT (CSF2RAWT) alleles (Fig. 3 A), whereas only the CSF2RAWT sequence was detected in the mother (Fig. 3 A). The sequence of CSF2RA in leukocyte messenger RNA (mRNA) was similar to the respective DNA sequence for each family member, even when cDNA was overamplified to detect potentially underrepresented mRNA sequences (unpublished data). This absence of a normal sequence for CSF2RA exon 7 in both siblings and the absence of the point mutation in the mother suggested that one maternal CSF2RA allele may be deleted in the patient, her sister, and their mother. The nucleotide sequence of CSF2RB in leukocyte mRNA from all family members was normal (unpublished data).
PCR amplification demonstrated a reduced copy number of CSF2RA in genomic DNA from both siblings and their mother compared with the father and a healthy control (Fig. S4). In contrast, the genomic copy number of CSF2RB was similar in all family members and a control (Fig. S4). Routine cytogenetic analysis was normal in all family members (Fig. S5), ruling out a large chromosomal abnormality. Fluorescent in situ hybridization (FISH) demonstrated a deletion in the pseudoautosomal region 1 of the X chromosome in the patient (Fig. 3 B), her sister, and mother (Fig. S6) that was not present in the father (Fig. 3 B). Further evaluation using comparative genomic hybridization (CGH) demonstrated an interstitial deletion encompassing the region from ∼1,610,183 to 2,873,864 bp on one maternal-derived X chromosome in the patient (Fig. 3 C), her sister, and mother (not depicted). High-resolution single-nucleotide polymorphism (SNP) mapping further defined this small deletion as extending from 1,308,324 to 2,881,011 bp in one maternal X chromosome (Fig. 3 D) and in both siblings (Fig. S7). Importantly, this deletion encompassed the CSF2RA allele (Fig. 3 E), thereby explaining the unexpected CSF2RA DNA and mRNA sequence results.
These results demonstrate the inheritance of compound heterozygous abnormalities of CSF2RA in both siblings (Fig. 3 F). The G174R mutation present on the paternal X chromosome, localized to the extracellular cytokine binding domain of GM-CSF-Rα, alters a potential site of glycosylation and impairs GM-CSF binding and receptor signaling. However, it is unclear if the functional impairment is caused by the change in primary amino acid sequence or loss of glycosylation.
Reproduction of the CSF2RAG174R abnormality
To reproduce the GM-CSF receptor signaling defect identified in this family, plasmids expressing CSF2RB and either CSF2RAG174R or CSF2RAWT were cotransfected into 293 cells. Western blotting showed receptor proteins of the expected size from CSF2RAWT and smaller molecular forms from CSF2RAG174R (Fig. 4 A), which is similar to results for primary leukocytes (Fig. 2 B). Treatment of transduced cell lysates with Peptide: N-Glycosidase F, an amidase that cleaves between the innermost GlcNAc and asparagine residues of high mannose oligosaccharides (3), resulted in a similar electrophoretic pattern for normal and G174R GM-CSF-Rα. This confirmed the G174R mutation alters N-linked glycosylation (Fig. 4 A) by disrupting a functional N-linked glycosylation site and that glycosylation is responsible for the smaller molecular forms from this allele. GM-CSF-Rβ was expressed in CSF2RB-transfected cells but was undetectable in nontransfected cells (unpublished data). GM-CSF binding studies demonstrated that cells expressing CSF2RAWT and CSF2RB rapidly bound GM-CSF and removed it from the media, whereas cells expressing CSF2RAG174R and CSF2RB or CSF2RB alone did not clear GM-CSF (Fig. 4 B). Cotransfection of CSF2RB together with CSF2RAWT, but not with CSF2RAG174R, resulted in expression of functional GM-CSF receptors as demonstrated by GM-CSF–dependent phosphorylation of STAT5 only with the former combination (Fig. 4 C). Incubation of CSF2RAG174R/CSF2RB-transduced cells with higher concentrations of GM-CSF in the range of those observed in the index patient gave results similar to those obtained with primary leukocytes and clearly demonstrated a modest dose-dependent increase in STAT5 phosphorylation, which is consistent with a partial rescue of the signaling abnormality (Fig. 4 D).
Clinical relevance of the CSF2RAG174R mutation
Our findings demonstrate that familial primary PAP in humans is caused by compound heterozygous abnormalities of CSF2RA, including loss of heterozygosity and a function-altering point mutation that severely reduces GM-CSF receptor signaling. These are the first results to demonstrate familial primary PAP in humans and that GM-CSF-Rα is critical for surfactant homeostasis. They extend previous findings that PAP is caused by the absence of GM-CSF or GM-CSF-Rβ in mice (4–6) or is specifically associated with the presence of neutralizing GM-CSF autoantibodies or a reduction in alveolar macrophage numbers or function in humans (9, 10). The allelic frequency of these mutations has not been determined but may be rare given that autoantibodies against GM-CSF account for most individuals with PAP.
Both affected siblings had impaired pulmonary function and failure to thrive, supporting the conclusion that prolonged respiratory insufficiency in children retards growth. Important to this conclusion is the observation that Xp22.33 deletion did not include the SHOX (short stature homeobox) gene, the deletion of which impairs growth (28). Further, heterozygous loss of CSF2RA in the mother did not impair growth. Early results suggest that whole lung lavage therapy was followed by an increase in growth acceleration in the index patient.
This study highlights the usefulness of several novel biomarkers in the diagnostic evaluation of PAP. The combination of a negative GM-CSF autoantibody test (29) and an abnormal CD11b stimulation index (24) predicted the GM-CSF receptor dysfunction, which was confirmed by the lack of GM-CSF binding and defective GM-CSF receptor signaling. GM-CSF levels were increased in lung lavage from the patient and, in serum from the patient and her sister, compared with family members and healthy controls. Because primary PAP is presumed to be caused by a defect in GM-CSF regulation of alveolar macrophage surfactant catabolism (9, 10), an elevated serum GM-CSF level may be useful as a biomarker to distinguish primary PAP caused by GM-CSF autoantibodies (90% of all cases of PAP) (9, 10, 13, 30) from PAP caused by GM-CSF receptor dysfunction (this study; references 14–16) and PAP caused by GM-CSF deficiency (4, 5) (not yet reported in humans). Finally, an increased serum SP-D level predicted the occurrence of PAP in the sister, which was confirmed by subsequent clinical and radiographical evaluation.
The slow progression and differences in severity of the lung disease between affected siblings with the same CSF2RA mutations and GM-CSF-Rα dysfunction indicates that penetrance is important in familial PAP because of GM-CSF receptor defects. The increased levels of GM-CSF likely reflect impaired clearance but are also consistent with interruption of a negative-feedback regulatory loop. Results strongly suggest that the high levels of pulmonary GM-CSF present in epithelial lining fluid result in a reduced but not abrogated receptor signaling and provide a molecular explanation the slow progression of PAP in both affected siblings. It is currently unclear if the level of increased pulmonary GM-CSF correlates with disease severity. They also suggest that aerosolized delivery of GM-CSF to the lungs may be therapeutic because this approach could theoretically achieve pulmonary GM-CSF levels several thousand–fold higher than endogenous levels measured in the patient and 20-fold higher than levels that appeared to completely restore GM-CSF receptor signaling in the transfection experiments. BM transplantation and CSF2RA gene transfer represent other potentially therapeutic approaches.
The identification of mild PAP in the older sister provides a unique and important opportunity to observe PAP at a very early stage of disease development. Finally, this case exemplifies the utility of a collaborative approach to the diagnosis and treatment of individuals with very rare diseases involving geographically disparate clinical centers and the benefit derived from integrating basic science, clinical medicine, and translational research.
Materials And Methods
The institutional review board of the Cincinnati Children's Hospital Medical Center approved the study, and all participants gave written informed consent or assent. Detailed case histories and laboratory data for the participants are included in the supplemental text .
GM-CSF receptor structure and function.
GM-CSF receptors were evaluated by flow cytometry using antibodies recognizing GM-CSF-Rα (BD) or β (Santa Cruz Biotechnology, Inc.) and by Western blotting (7) with chain-specific antibodies (Santa Cruz Biotechnology, Inc.). The effects of glycosylation on GM-CSF-Rα were evaluated by incubating cell lysates with PNGase F (New England Biolabs, Inc.) (3). Actin was measured as a loading control using anti-actin antibodies (Santa Cruz Biotechnology, Inc.). To evaluate receptor-mediated clearance of GM-CSF, Ficoll-purified blood leukocytes or 293 cells (American Type Culture Collection) were cultured in DMEM containing 10% fetal bovine serum and human GM-CSF (Leukine; Bayer) at 1 ng/ml. At subsequent times, GM-CSF was measured in the culture medium by ELISA (R&D Systems) (24). To evaluate GM-CSF receptor signaling, heparinized blood was incubated with or without 10 ng/ml of human GM-CSF Leukine for 15 min. After lysing RBC, leukocytes were evaluated by Western blotting (24) using anti–human STAT5 (Santa Cruz Biotechnology, Inc.) and anti–phospho-STAT5 (Millipore) antibodies. The CD11b stimulation index was measured as previously described (24).
SP-D, GM-CSF, MCP-1 (monocyte chemotactic protein-1), M-CSF, and urea were measured using commercial ELISA kits (BioVendor, R&D Systems, BD, R&D Systems, and BioAssay Systems, respectively) (24). The concentration of cytokines in epithelial lining fluid of the lungs was estimated by multiplying the concentration in lavage fluid by the ratio of urea concentrations in serum and lavage fluid (26).
Genomic DNA or total mRNA was purified from blood leukocytes with commercially available kits (QIAGEN). The PCR was used to generate products spanning exons and flanking intronic sequences (genomic DNA) or the coding sequences and 5′ and 3′ flanking untranslated sequences (mRNA) of the CSF2RA and CSF2RB genes, which were then analyzed by direct sequencing. The resulting sequences were compared with published sequences (from GenBank/EMBL/DDBJ under accession nos. NM_006140.3 and BC070085, respectively). The relative copy number of CSF2RA and CSF2RB genes among family members and a healthy control was measured by amplification of genomic DNA using PCR with gene-specific oligonucleotides followed by analysis of products by electrophoresis on 2% agarose gels. Experimental methods for measuring gene copy number, evaluating the genes encoding SP-B, SP-C, and ABCA3 (19, 31) for function-altering mutations, cytogenetic analysis, CGH analysis, FISH, and high-resolution SNP array analysis (32) are described in the supplemental Materials and methods .
Reproduction of the GM-CSF receptor defect.
The methods used to clone and express mRNA transcripts encoding GM-CSF-Rα and β chains in cultured cells and evaluate protein glycosylation are described in the supplemental Materials and methods.
Numeric data are presented as means ± SE. Statistical comparisons were made with Student's t test for two-group comparisons and with one-way analysis of variance with posthoc analysis according to the Holm-Sidak method for multiple-group comparisons. P-values of <0.05 were considered to indicate statistical significance.
Online supplemental material.
Fig. S1 shows normalized growth curves for the index patient and her sister. Fig. S2 shows GM-CSF binding and clearance by blood leukocytes from affected and unaffected family members and control. Fig. S3 shows increased pulmonary GM-CSF levels in GM-CSF receptor β chain (Csf2rb) KO mice. Fig. S4 shows evaluation of CSF2RA and CSF2FB gene copy number by PCR amplification. Fig. S5 shows karyotypes of high-resolution GTG-banded chromosomes. Fig. S6 shows evaluation of heterozygosity at the CSF2RA locus by FISH. Fig. S7 shows high-resolution SNP analysis. The supplemental text contains additional details regarding the case histories and clinical laboratory and other data cited throughout the text.
© 2008 Suzuki et al. This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.jem.org/misc/terms.shtml). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 3.0 Unported license, as described at http://creativecommons.org/licenses/by-nc-sa/3.0/).
We thank Diana Koch, Diane Black, Paula Blair, and Ralph Stanley (Cincinnati Children's Hospital Medical Center) for excellent technical assistance, Dr. Denis McGraw (University of Cincinnati Medical Center) for performing bronchoalveolar lavage in healthy controls, Dr. Frank McCormack (University of Cincinnati College of Medicine) for helpful discussions, and Lauren Vannoy (Wake Forest University School of Medicine and the Translational Research Trials Office, Cincinnati Children's Hospital Medical) for clinical research coordination.
This work was supported in part by the grants from the National Center For Research Resources (RR019498 to B.C. Trapnell to support the Rare Lung Diseases Consortium) and the National Heart, Lung, and Blood Institute (HL085453 to B.C. Trapnell and HL085610 to J.A. Whitsett), and the Division of Pulmonary Biology at the Children's Hospital Medical Center.
The authors have no conflicting financial interests.